It's easier to claim that a real adblocker is a net negative for performance while implementing in a slow language than admit that they can't realistically do better this decade due to the complexity of the task and the fact that it's basically one guy's part time project.